<<

Has the Effectiveness of Changed? Impact of the Variety of Lithium’s Effects Paul Grof, M.D., Ph.D.

Lithium treatment, initially considered specific for bipolar accumulation of atypical, treatment-resistant patients in disorder, has since been shown to provide additional benefits academic centers and, in particular, the broadening of in affective and other disorders. This variety of benefits diagnoses of affective disorders, further complicate the should be taken into account when interpreting recently interpretation of the recent reports. Lithium, however, reported lower efficacy during lithium prophylaxis, as well continues working well for patients with typical bipolar as early relapses and loss of efficacy after lithium disorders, for whom it was originally proved effective. discontinuation. There are particularly striking parallels [Neuropsychopharmacology 19:183–188, 1998] between these recent reports and earlier observations of © 1998 American College of Neuropsychopharmacology. “” lithium effects. Other factors, such as the Published by Elsevier Science Inc.

KEY WORDS: Lithium effects; Prophylactic effect; that when bipolar and unipolar patients stabilized on Antipsychotic effect; Lower efficacy of lithium; lithium were randomly allocated to lithium and pla- Discontinuation of lithium; Loss of efficacy cebo, only the placebo patients subsequently relapsed. It has been nearly 50 years since (1949) first In addition to its demonstrated efficacy, lithium was reported on the specific treatment of acute with also thought to be a specific drug for manic-depressive lithium salts and about 25 years since lithium was illness. Cade was so struck by the specificity that he en- widely accepted as an effective long-term treatment. tertained the possibility of lithium deficiency in his pa- When lithium treatment first achieved wide acceptance, tients. Later, the conviction about specificity was drawn it was welcomed as a treatment that had the best dem- mainly from the initial observations that lithium could onstrated efficacy among psychiatric treatments. A se- prevent recurrences of primary affective disorders but ries of double-blind as well as large open studies (as re- not of other psychiatric conditions, and at the same viewed, for example, in Schou and Thomsen 1975; time, had only negligible effects on normal moods. Schou 1982) demonstrated uniformly lithium’s useful- Lithium seemed to be a substance with a specific bene- ness, as well as its superiority over a placebo and anti- fit, when compared with the broad spectrum of applica- depressants in the long-term treatment of affective dis- tions of other psychotropic agents. For example, tricy- orders. The double-blind discontinuation study of clic antidepressants could treat conditions as varied as Baastrup et al. (1970) presented particularly robust mental , chronic pain, and nocturnal enure- findings, unparalleled in , which showed sis. This tenet of lithium’s specificity was so mighty that clinicians often tended to conclude ex iuvantibus; when patients treated with lithium benefited, it was resolved that they must be suffering from some form of bipolar From the Department of Psychiatry, University of Ottawa, Royal illness. In hindsight, it would have been more produc- Ottawa Hospital, 1145 Carling Avenue, Ottawa, Ontario, . tive to question lithium’s absolute specificity. Address correspondence to: P. Grof, M.D., Ph.D., Department of Psychiatry, University of Ottawa, Ottawa, ONT K1Z 7K4, Canada. Over the years, however, the initially narrow scope Received February 23, 1998; accepted February 23, 1998. of lithium’s caliber in affective disorders has quickly ex-

NEUROPSYCHOPHARMACOLOGY 1998–VOL. 19, NO. 3 © 1998 American College of Neuropsychopharmacology Published by Elsevier Science Inc. 0893-133X/98/$19.00 655 Avenue of the , New York, NY 10010 PII S0893-133X(98)00023-2

184 P. Grof NEUROPSYCHOPHARMACOLOGY 1998–VOL. 19, NO. 3

panded, adding to its well-established antimanic and quency. Second, both expected and expected prophylactic benefits antiaggressive effects (particu- depressions were significantly reduced. Third, the pro- larly studies of Sheard 1978, 1984), antipsychotic poten- phylactic effect was dependent upon therapeutic lith- tial (e.g., investigations of Garver et al. 1988; Garver ium concentrations, and toxicity was observed only Hutchinson 1988; Lenz et al. 1987, Lenz et al. 1989), an- with elevated lithium levels. Fourth, when lithium tisuicidal and mortality-reducing ability (specifically, treatment was discontinued in remission, subsequent studies of Müller-Oerlinghausen et al. 1992; Coppen et recurrences developed gradually, and no rebound was al. 1992), and possibly also antidepressant (Mendels observed. The benefit was generally reproducible by re- 1976, 1982) and antidepressant-augmenting effects (de instituting lithium treatment. Montigny et al. 1983). Surprisingly, while quickly expanding therapeutic Antimanic Effect horizons, lithium has also been reported to be gradually less and less effective (Grof et al. 1993; Guscott and Tay- The value of lithium as an antimanic drug has been well lor 1994), particularly in clinical practice. Recent litera- established against a placebo and neuroleptics; there- ture certainly paints a rather pessimistic picture; it fore, lithium is now considered the standard antimanic seems now that only a fraction of patients actually ben- drug. Similar to the current controversy about lithium’s efit from lithium prophylaxis (Prien and Gelenberg prophylactic efficacy, recent findings about the anti- 1989), and a sizeable proportion of patients may lose manic effects of lithium have been less uniform, some the benefit altogether over time (Post et al. 1992). Fur- confirming (Bowden et al. 1994) and others questioning thermore, discontinuation of lithium is often followed lithium’s superiority over placebo. The issue of whether by a prompt recurrence of the illness (Suppes et al. the prophylactic and antimanic benefits are of the same 1991), and later possibly by a loss of efficacy (Post et al. nature and exert themselves in the same patients has 1992). In brief, recent literature suggests that although not yet been systematically investigated. Clinically, lithium treatment does not help much anymore, after its there seems to be an association. However, the anti- discontinuation, many patients may now suffer even manic effect at times seems to act with a wider range, as more. if there were at least two possible mechanisms in- Thus, during the past two decades, the earlier per- volved: a relatively specific effect in bipolar mania; and ception of lithium as an effective and specific treatment a nonspecific, overactivity-reducing effect in some has been changing from that of a highly valued, long- pathological states of other provenience. Further under- term treatment to a questionably useful substance of standing of lithium’s antimanic activity may surface fleeting benefit. It is important to explore the nature of from the emerging comparative studies with such anti- this shift and the possible misunderstandings that sur- epileptics as valproic acid and . round it. We suggest that a combination of several fac- tors has contributed to this controversy, in particular a Antiaggressive Effect lack of recognition that lithium treatment offers more than one distinct clinical benefit; a broadening of the di- The antiaggressive effects of lithium have been demon- agnoses of affective disorders; the need to differentiate strated amply in a series of studies, in particular, between efficacy and effectiveness; and an accumula- Sheard (1978, 1984). The effects were documented in tion of difficult-to-treat affective disorders in academic several different populations: psychiatric patients, men- research centers. tally retarded populations, and penitentiary subjects. Sheard and other investigators who studied the antiag- gressive effects of lithium argued convincingly that VARIETY OF LITHIUM EFFECTS subjects who benefited did not suffer from bipolar ill- ness: the effect was distinctly different. A variety of distinct effects of lithium treatment have been described, along with their clinical characteristics. “Antipsychotic” Effect In terms of lithium’s specificity, probably one of the “Prophylactic,” Normal Mood Stabilizing Effect most perplexing effects of lithium has been its “antipsy- The important clinical characteristics of this most chotic” capacity. Both acute and long-term “antipsy- widely utilized, well-documented benefit have been chotic” effects have been described best by Garver and outlined, in particular, in the large series of the studies his (Garver et al. 1984, 1988) in of lithium in the late 1960s and early 1970s. First, the re- and schizophreniform psychoses, and illustrated on currences of abnormal moods were significantly re- schizoaffective psychoses by many others (e.g., Angst et duced in about three-quarters of the patients, either al. 1970; Lenz et al. 1987; Lenz et al. 1989) and can com- suppressed completely or reduced substantially in fre- monly be seen clinically. During treatment with lithium

NEUROPSYCHOPHARMACOLOGY 1998–VOL. 19, NO. 3 Has Effectiveness of Lithium Changed? 185

alone, a dramatic clearing of psychotic symptoms has edly reduces mortality and suicidal behavior and that been seen in acute psychoses (schizoaffective and schizo- the mortality of affective disorders is diminished to a phreniform), even in mood incongruent ones, and in level indistinguishable from that of the general popula- many patients, continued treatment with lithium alone tion (e.g., Müller-Oerlinghausen et al. 1992, Coppen et could prevent further episodes. Garver et al., for example, al. 1992). The antisuicidal effect is also supported by treated a large consecutive series of mood incongruent findings from a comparative, long-term study of lith- psychotic episodes with lithium only. These patients were ium and , in which only lithium-treated diagnosed as having DSM-III affective, mood-incongruent patients remained completely free of suicides (Greil et psychoses, schizoaffective disorders, schizophreniform al. 1997). In patients with a history of suicide attempts, disorders, and schizophrenias. None had received depot a reduction was observed both in responders and non- neuroleptics within 6 weeks of admission. Each patient responders to lithium treatment, indicating that this underwent a systematic trial with lithium distinct antisuicidal benefit takes place over and above alone, with levels in the range of 1.0 to 1.2 mEq. lithium’s prophylactic effect. Excellent responses with remission were seen in 29% of patients. Lithium responders fared well, without the in- Antidepressive Effect and Lithium Augmentation troduction of neuroleptics at any time during the course of hospitalization and could be discharged on lithium The antidepressive effect of lithium was first reported alone, essentially symptom free. Of the lithium-respon- by Vojtechovsky and later documented (in particular, sive patients who were readmitted, about half showed a Mendels 1976, 1982). This effect can be prompt and concordant lithium response on readmission. Unlike the quite marked in depressed patients who suffer from generally reproducible response to lithium in typical af- some form of bipolar illness, but it has not yet been fective disorders, in only half of these atypical patients studied systematically enough. There seems to be clini- could the response be replicated. Using the terminology cal agreement that the antidepressive effect of lithium is of recent years, we could say that the initially well-proved infrequent, which is in contrast with the commonly ob- response “was lost” in one-half of the patients. served antimanic effect. This “antipsychotic” effect of lithium superficially Since 1981, de Montigny et al. (e.g. 1983) have devel- resembles both the prophylactic and antimanic effects, oped the concept of augmenting antidepressants by although closer observation exposes a different profile. adding lithium. The idea of such a combination has been First, patients benefiting from the antipsychotic effect of supported by other observations in the literature and lithium experience a reduction of manias but no reduc- found clinically useful, although the evidence for a tion in their depressive episodes (Grof 1994). Second, “rapid augmentation” (acting within 48 hours) seems less when these patients are readmitted for mania, the anti- clear. One of the interesting features of augmentation manic effect is reproducible in only about half of the pa- is that it may work even in the absence of therapeutic tients (Garver et al. 1988; Garver and Hutchinson 1988), lithium levels. Thus, this clinically useful phenomenon as if there had been a loss of efficacy over time or a loss may require additional investigations of the underlying of efficacy on discontinuation. It would seem that, for a mechanisms. satisfactory long-term effect, in about half of these ini- tial responders, lithium treatment must eventually be Implications combined with, or replaced by, other psychotropic drugs. Third, may develop, even with Lithium may possibly still be considered more specific therapeutic lithium levels. Fourth, when lithium is dis- than other psychotropic drugs; however, it clearly ex- continued, patients with atypical affective or mood- hibits more than one clinically distinct, qualitatively incongruent disorders frequently present with early re- different effect. To wit, prophylactic effect (stabilizing lapses (Lenz et al. 1988, Grof 1994). normal mood) and “antipsychotic” effect are each These observations strongly suggest that the prophy- linked with different clinical characteristics. Some of the lactic, normal mood-stabilizing effect in typical affec- effects may have a common denominator in improved tive disorders and the “antipsychotic” effect in atypical serotonergic function, but the clinical effects have dis- disorders are two different benefits, with possibly con- tinguishing features and may reflect different mecha- nected, but different, underlying mechanisms. It may nisms of action and differences in the pathophysiology be prudent to consider these two lithium benefits sepa- of treated conditions. It would be prudent to study rately, both at the clinical and the basic science level. these phenomena separately for each effect, in particu- lar for the prophylactic and “antipsychotic” effects, and separately in patients with typical affective disorders Reduction of Suicidal Behavior and Mortality for whom the efficacy of lithium was originally proved. There now exists a large body of data compatible with Averaging from heterogeneous groups can only blur the conclusions that long-term lithium treatment mark- any findings.

186 P. Grof NEUROPSYCHOPHARMACOLOGY 1998–VOL. 19, NO. 3

So far, we have focused on the variety of benefits disorder will be considered appropriate for lithium pro- lithium administration can offer. The other factors that phylaxis. Making an accurate diagnosis and estimating confound the evaluation of lithium’s prophylactic value recurrence risk becomes much more challenging. Thus, are mentioned only briefly, because they have already as a result of the shorter observation combined been described elsewhere: the need to differentiate be- with an expanded diagnostic approach, many patients tween efficacy and effectiveness; accumulation of treat- who would not have been placed on lithium 25 years ment-resistant patients in academic centers; and broad- ago are now entering clinical investigations. An espe- ening of the diagnosis of affective disorders. cially confusing element of this tendency has been the addition of the mood-incongruent psychoses to affec- tive disorders in DSM-III. DIFFERENTIATING BETWEEN EFFICACY As we have shown recently (Grof et al. 1995), com- AND EFFECTIVENESS pared to 25 years ago, the patients included in many re- cent studies of affective disorder differ not only in lower When interpreting the reports on the low success of treatment responsiveness, but in many other character- lithium , it is very important to keep in mind the istics as well: pre-existing psychopathology; age at onset; difference between evaluating efficacy and effective- clinical course; comorbidity; and so forth. For example, ness, as Guscott and Taylor (1994) tellingly related. in family studies, the proportion of relatives diagnosed When evaluating efficacy in the initial clinical trials, pa- with affective disorders has increased five to six times tients are selected according to strict criteria. The proce- over the past 20 years, and a review indicates that this in- dure should optimize detection of the putative efficacy crease may have taken place at the expense of other di- of the evaluated new substance. This is neither the case agnostic categories. Clearly, the recent studies deal with in clinical practice, nor in clinical trials during the later different populations of affective disorders, diagnosed stages. more broadly than 25 years ago, including those in whom prophylactic lithium effects were not demonstrated.

ACCUMULATING TREATMENT-RESISTANT DISCUSSION PATIENTS Since the original studies of lithium in the late 1960s Academic centers, which have become the main per- and early 1970s, several noteworthy changes have oc- former of clinical trials, tend to accumulate more than curred. In the wider use of lithium, a variety of other their share of difficult, atypical, treatment-resistant pa- benefits has been described. Opinions may vary about tients with affective disorders who then enter clinical the actual specificity of lithium, but such effects as pro- trials (e.g., Gershon 1996 and other investigators). phylactic, antiaggressive, and antipsychotic exhibit dis- tinct clinical features. Furthermore, for several reasons, the populations researched and treated with lithium to- BROADENING THE DIAGNOSIS OF day differ from those for whom lithium was proved ef- AFFECTIVE DISORDER fective. As a result, the earlier perception of lithium as an effective and specific treatment has been markedly chang- There is now substantial evidence that during the past ing. If correct, this mutation would be a sad state of af- two decades, affective disorders have been diagnosed fairs — rather tragic for clinical practice, confusing for more liberally. In the 1950s, during the days of enthusi- research, and certainly, a source of medicolegal problems, astic of , there was concern because lawyers now bring suits for either prescribing about the “disappearing manic depressive.” Conversely, lithium, or discontinuing it or not prescribing it. during the fervent use of lithium and antidepressants, Parenthetically, clinical and medicolegal concerns investigators (e.g., Baldessarini 1970; Grof and Fox about tardive dyskinesia have often led to overempha- 1987; Stoll et al. 1993) documented a strong opposite sizing mood abnormalities and treating patients with trend favoring the diagnoses of affective disorders, at lithium in place of, or in combination with, a low dose the expense of other psychiatric conditions. neuroleptic. In some jurisdictions, a psychiatrist now Because of different circumstances, 25 years ago, pa- must provide special justification for prescribing a neu- tients were placed on long-term lithium according to roleptic but not for such other psychotropic drugs as narrower diagnostic criteria and after a much longer lithium. As recent statistics indicate, lithium has been period of diagnostic observation. For example, before given, alone and in combination, to a majority of inpa- lithium treatment, Baastrup and Schou’s (1967) patients tients in several U.S. hospitals (Baldessarini et al. 1995). experienced nine episodes of illness on the average. At Such statistics make sense only if lithium is given present, many patients with only two episodes of mood widely outside of its proved indications.

NEUROPSYCHOPHARMACOLOGY 1998–VOL. 19, NO. 3 Has Effectiveness of Lithium Changed? 187

While we are exploring what seem to be new phe- Baastrup PC, Schou M (1967): Lithium as a prophylactic nomena in lithium prophylaxis — markedly reduced agent. Its effect against recurrent depressions and manic- long-term efficacy, and after discontinuation, both early depressive . Arch Gen Psychiat 16:162–172 relapses and a frequent loss of efficacy — it may be use- Baastrup PC, Poulsen JC, Schou M, Thomsen K, Amdisen A ful to keep in mind that several important ingredients (1970): Prophylactic lithium: Double-blind discontinua- tion in manic-depressive and recurrent-depressive dis- have changed since the initial studies, and phenomena orders. Lancet 2:326–330 similar to the newly reported ones have already been Baldessarini RJ (1970): Frequency of diagnosis of schizophre- observed in connection with the antipsychotic effect of nia versus affective disorders from 1944 to 1968. Am J lithium. These new phenomena are real, but the ques- Psychiat 127:759–763 tion remains whether they are expandable to those pa- Baldessarini RJ, Kando JC, Centorrino F (1995): Hospital use tients for whom lithium was demonstrated as effective of antipsychotic agents in 1989 and 1993. Am J Psychiat prophylactic treatment more than two decades ago. 152:1038–1044 There are parallels between the new observations Berghöfer A, Müller-Oerlinghausen B, Ahrens B (1996): Loss and some of the features of the earlier described “anti- of efficacy after discontinuation? Paper presented at the psychotic” effects of lithium. Naturally, a parallel is not International Lithium Conference, Malta, November a proof; however, these analogies between the recent 1995 findings about lithium prophylaxis and the earlier re- Bowden CL, Brugger AM, Swann AC, Calabrese JR, Janicak ports in atypical and schizophreniform disorders are PG, Petty F, Dilsaver SC, Davis JM, Rush AJ, Small JG, rather striking; in both cases only less than 30% of pa- Garza-Trevino ES, Risch SC, Goodnick PJ, Morris DD (1994): Efficacy of divalproex vs. lithium and placebo in tients benefited from prophylaxis (22% Prien and Ge- the treatment of mania. JAMA 271:918–922 lenberg 1989 vs. 27–29% Garver et al. 1988); in both Cade J (1949): Lithium salts in the treatment of psychotic cases, there is no beneficial effect on depression (Prien excitment. Med J Australia 2:349–352 and Gelenberg 1989; Grof 1994) after lithium discontin- Coppen A, Bailey J, Houston B, Silcocks P (1992): Lithium uation, a rebound and early relapses have been re- and mortality: A 15-year follow-up. Clin Neuropharma- ported often in both (e.g., Lenz et al. 1988, 1989; Suppes col 15:448A–449A et al. 1991; Grof 1994), and, finally, in both groups, after de Montigny C, Cournoyer G, Morisette R, Langlois R, lithium discontinuation, reintroducing lithium has been Cailler G (1983): addition in tricyclic helpful only in about half the cases (46% Post et al. 1992 antidepressant-resistant unipolar depression: Correla- vs. 50% Garver et al. 1988). tions with the neurobiologic actions of tricyclic antide- Lithium continues working well for patients for pressant drugs in lithium on the serotonin system. whom it was proved effective (Berghöfer et al. 1996; Arch Gen Psychiatry 40:1327–1333 Greil et al. 1994). In a recently completed study, when Garver DL, Hirschowitz J, Fleishmann R, Djuric PE (1984): patients were carefully divided into typical (unipolar Lithium response and psychoses: A double-blind, pla- cebo-controlled study. Psychiatry Res 12:57–68 and bipolar) and atypical (schizoaffective), lithium was clearly prophylactically superior to carbamazepine in Garver DL, Hutchinson LJ (1988): Psychosis, lithium- induced antipsychotic response, and seasonality. Psy- bipolar patients and to amitriptyline in unipolar ones, chiatry Res 26:279–286 while being poorly effective, less so than carbam- Garver DL, Kelly K, Fried KA, Magnusson M, Hirschowitz J azepine, in schizoaffective disorders (Greil et al. 1994). (1988): Drug response patterns as a basis of nosology for Furthermore, in typical affective disorders treated in the mood-incongruent psychoses (the schizophrenias). specialized clinics, there was no evidence of a loss of ef- Psycholog Med 18:873–885 ficacy after discontinuation (e.g., Berghöfer et al. 1996) Gershon S (1996): Comments at the John Cade Symposium, and no evidence for rebound (Grof 1994). We must as- 20th CINP, Melbourne, Australia sume, however, that if typical and atypical patients Greil W, Ludwig-Mayerhofer W, Erazo N, Schochlin C, were mixed in one study, the result would depend pri- Schmidt S, Engel RR, Czernik A, Giedke H, Müller-Oer- marily on the proportion of each group. Thus, the linghausen B, Osterheider M, Rudolf GA, Sauer H, trends mentioned above pose a considerable challenge Tegeler J, Wetterling T (1997): Lithium versus carba- for the interpretation of research findings and for the mazepine in the maintenance treatment of bipolar disor- ders—a randomized study. J Affect Disord 43:151–161 development of useful guidelines for clinical practice. Grof P, Fox D (1987): Admission rates and lithium therapy. Brit J Psychiat 150:264–265 Grof P, Alda M, Grof E, Fox D, Cameron P (1993): The chal- lenge of predicting response to stabilizing lithium treat- REFERENCES ment: The importance of patient selection. Brit J Psychiat 163:16–19 Angst J, Weis P, Grof P, Baastrup PC, Schou M (1970): Lith- Grof P (1994): Lithium discontinuation in typical and atypi- ium prophylaxis in recurrent affective disorders. Brit J cal affective disorders. Abstracts, 19th CINP, Washing- Psychiat 116:604–614 ton, June 1994

188 P. Grof NEUROPSYCHOPHARMACOLOGY 1998–VOL. 19, NO. 3

Grof P, Alda M, Ahrens B (1995): Clinical course of affective with manic-depressive illness and schizoaffective ill- disorders. Were and Jules Angst wrong? ness. Acta Psychiat Scand 86:218–222 Psychopathology 28:73–80 Post RM, Leverich GS, Altschuler L, Mikalauskas K (1992): Guscott R, Taylor L (1994): Lithium prophylaxis in recurrent Lithium discontinuation-induced refractoriness: Prelim- affective illness. Efficacy, effectiveness, and efficiency. inary observations. Am J Psychiat 149:1727–1729 Br J Psychiat 164:741–746 Prien R, Gelenberg AJ (1989): Alternatives to lithium for the Lenz G, Lovrek A, Thau K, Topitz A, Denk A, Simhandl C, preventative treatment of . Am J Psy- Wancata J, Wolf R (1987): Lithiumprophylaxis of schizo- chiat 146:840–848 affective disorders (in German). Wien Med Wochenschr Schou M, Thomsen K (1975): Lithium prophylaxis of recur- 137:5 rent endogenous affective disorders. In Johnson FR (ed), Lenz G, Lovrek A, Thau K, Topitz A, Denk A, Simhandl C, Lithium Research and Therapy. New York, Academic Wancata J, Wolf R (1988): Lithium withdrawal study in Press, pp 3–85 schizoaffective patients. In Birch N J (ed), Lithium; Inor- Schou M (1982): Trends in lithium treatment and research ganic Pharmacology, and Psychiatric Use. Oxford, IRL during the last decade. Pharmacopsychiatria. 15:128–130 Press Ltd Sheard MH (1978): The effect of lithium and other on Lenz G, Wolf R, Simhandl C, Topicz A, Berner P (1989): aggressive behavior. Mod Prob Pharmacopsychiatry Long-term progosis and recurrence prophylaxis of 13:53–68 schizoaffective psychoses (in German). In Marneros, A (ed), Schizoaffektive Psychosen (Tropon Symposium, Sheard MH (1984): Clinical pharmacology of aggressive vol IV). Berlin, Springer, pp 55–66 behavior. Clin Neuropharmacol 7:173–183 Stoll AL, Tohen M, Baldessarini RJ, Goodwin DC, Stein S, Katz Mendels J (1976): Lithium in the treatment of depression. S, Greenens D, Swinson RP, Goethe JW, McGlahan T Am J Psychiatr 133:373–378 (1993): Shift in diagnostic frequencies of schizophrenia and Mendels J (1982): Role of lithium as an antidepressant. Mod major affective disorders at six North American psychi- Prob Pharmacopsychiat 18:138–144 atric hospitals 1972–1988. Am J Psychiat 150: 1668–1673 Müller-Oerlinghausen B, Ahrens B, Grof E, Grof P, Alda M, Suppes T, Baldessarini RJ, Faedda GL, Tohen M (1991): Risk of Schou M, Thau K, Wolf R, Lenz G (1992): The effect of recurrence following discontinuation of lithium treatment long-term lithium treatment on the mortality of patients in bipolar disorder. Arch Gen Psychiat 48:1082–1088